A project that uses word frequencies to determine the author of a given text snippet.

A simple histogram of word frequencies can be used to help determine authorship of a piece of text. This program provides a (slightly) more sophisticated method where n-grams of text map to a set of possible subsequent words where each word has an associated frequency.

Consider the following example text:

This is an example text.  This is the example.

Using an n-gram size where n = 2 where we process the following mappings:

(This, is) -> [an, the]
(is, an) -> [example]
(an, example) -> [text]
(example, text.) -> [This]
(text., This) -> [is]
(is, the) -> [example.]

The phase "This is" is equally likely to continue with either "an" or "the". This mapping provides a crude encoding of the author's style and can be used to help identify authorship if samples of the author's texts are available for comparison.